Estimation based on hybrid censored data from the power Lindley distribution

We consider classical and Bayesian procedures for estimating the parameters of the power Lindley distribution based on hybrid censored data. By unifying the likelihood function under the hybrid censoring scheme, we consider the maximum likelihood estimators (MLEs) and develop an expectation-maximization algorithm to find the MLEs. We adopt the asymptotic distribution of the MLEs to construct approximate confidence intervals. We then study Bayesian estimators and the credible intervals of the parameters under appropriate choices of prior distributions. Lindley’s approximation method and Markov chain Monte Carlo sampling algorithm are also provided to evaluate these Bayesian estimators. The finite sample performances of these estimation methods are investigated using simulation studies and a real data example.